identification of novel celsr1 mutations in spina bifida

Identification of Novel CELSR1 Mutations in Spina BifidaYunping Lei1, Huiping Zhu1, Wei Yang3, M. Elizabeth Ross4, Gary M. Shaw3, Richard H. Finnell1,2*

1 Dell Pediatric Research Institute, Department of Nutritional Sciences, The University of Texas at Austin, Austin, Texas, United States of America, 2 Department of

Chemistry, College of Natural Sciences, The University of Texas at Austin, Austin, Texas, United States of America, 3 Department of Pediatrics, Division of Neonatology,

Stanford University School of Medicine, Stanford, California, United States of America, 4 Center for Neurogenetics, Brain and Mind Research Institute, Weill Cornell Medical

College, New York, New York, United States of America

Abstract

Spina bifida is one of the most common neural tube defects (NTDs) with a complex etiology. Variants in planar cell polarity(PCP) genes have been associated with NTDs including spina bifida in both animal models and human cohorts. In this study,we sequenced all exons of CELSR1 in 192 spina bifida patients from a California population to determine the contribution ofCELSR1 mutations in the studied population. Novel and rare variants identified in these patients were subsequentlygenotyped in 190 ethnically matched control individuals. Six missense mutations not found in controls were predicted to bedeleterious by both SIFT and PolyPhen. Two TG dinucleotide repeat variants were individually detected in 2 spina bifidapatients but not detected in controls. In vitro functional analysis showed that the two TG dinucleotide repeat variants notonly changed subcellular localization of the CELSR1 protein, but also impaired the physical association between CELSR1 andVANGL2, and thus diminished the ability to recruit VANGL2 for cell-cell contact. In total, 3% of our spina bifida patients carrydeleterious or predicted to be deleterious CELSR1 mutations. Our findings suggest that CELSR1 mutations contribute to therisk of spina bifida in a cohort of spina bifida patients from California.

Citation: Lei Y, Zhu H, Yang W, Ross ME, Shaw GM, et al. (2014) Identification of Novel CELSR1 Mutations in Spina Bifida. PLoS ONE 9(3): e92207. doi:10.1371/journal.pone.0092207

Editor: Osman El-Maarri, University of Bonn, Institut of experimental hematology and transfusion medicine, Germany

Received October 2, 2013; Accepted February 20, 2014; Published March 14, 2014

Copyright: � 2014 Lei et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricteduse, distribution, and reproduction in any medium, provided the original author and source are credited.

Funding: This work was supported in part by NIH grants HD067244, NS076465, and ES021006. The funders had no role in study design, data collection andanalysis, decision to publish, or preparation of the manuscript.

Competing Interests: The authors have declared that no competing interests exist.

* E-mail: [email protected]

Introduction

Neural tube defects (NTDs) are among the most severe and

common of all human birth defects. The most frequent types of

NTDs are spina bifida and anencephaly. The etiology of NTDs is

complex and involves both environmental and genetic factors.

Periconceptional folic acid supplementation reduces 50% to 70%

of newborn NTDs [1]; however, the mechanisms underlying this

protective effect remain unclear. In terms of genetic underpin-

nings, monozygotic twinning and single gene disorders have long

been associated with increased risks of NTDs [2]. Numerous

exploratory candidate gene studies have highlighted a variety of

biological pathways such as the folate and one carbon metabolism

and transport [3], DNA repair [4], retinoic acid receptors [5], and

the planar cell polarity (PCP) signaling network [6].

The PCP pathway controls the polarity of cells within the plane

of epithelium in both vertebrates and invertebrates. The PCP

genes, including frizzled, dishevelled, vangl, flamingo (Celsr), prickle and

diego, were initially identified in Drosophila, and are highly

conserved throughout evolution [7]. PCP signaling is required

for the initiation of neural tube closure in higher vertebrates [8]. In

mice, mutations in Vangl2, Celsr1, Dishevelled and Frizzled result in

the NTD known as craniorachischisis [9]. In humans, mutations in

FRIZZLED6 [10], DISHEVELLED [11], VANGL [12,13] SCRIB

[14] and PRICKLE1 [15] have been associated with NTDs.

CELSR1 is well known for regulating the establishment and

maintenance of planar cell polarity. During mitosis, CELSR1

recruits VANGL2 and FZD6 to endosomes. Following mitosis,

CELSR1, VANGL2 and FZD6 are recycled to the cell surface to

re-establish cell polarity [16]. In Drosophila, CELSR1 mediates

homotypic interaction between adjacent cells and transmits

instructive PCP signals. In zebrafish, the knocking down of Celsr1

produced convergent extension (CE) defects [17]. In mice, Celsr1

mutants exhibited craniorachischisis, the most severe form of

NTDs [18]. In humans, functional CELSR1 single nucletotide

variants (SNVs) have been identified in fetuses with craniorachis-

chisis, and predicted-to-be-deleterious CELSR1 SNVs have been

detected in a few cases with NTDs or caudal agenesis [19,20].

However, the contribution of CELSR1 mutations in the etiology of

spina bifida is still unknown. In this study, we investigated the

CELSR1 coding region sequence among a cohort of spina bifida

infants born in California by Sanger sequencing. Further, we

conducted in vitro functional analyses to validate the functional

effect of the identified mutations.

Materials and Methods

Ethics statementThe approval process includes detailed review by the State of

California Committee for the Protection of Human Subjects (the

primary IRB). All samples were obtained with approval from the

State of California Health and Welfare Agency Committee for the

Protection of Human Subjects.

Human subjectsData were obtained from a population-based case–control study

conducted by the California Birth Defects Monitoring Program

(CBDMP). The CBDMP is an active, population-based surveil-

PLOS ONE | www.plosone.org 1 March 2014 | Volume 9 | Issue 3 | e92207

http://creativecommons.org/licenses/by/4.0/

Ta

ble

1.

CEL

SR1

rare

no

nsy

no

nym

ou

sva

rian

tsd

ete

cte

din

NT

Ds

bu

tn

ot

inco

ntr

ols

.

Nu

cle

oti

de

cha

ng

ers

IDa

ach

an

ge

Po

lyP

he

nS

IFT

Mu

tati

on

Ta

ste

rF

AT

HM

Ma

EV

Pb

1K

GP

cP

he

no

typ

eD

om

ain

c.3

06

8C

.G

NA

dp

.Ala

10

23

Gly

po

ssib

lyd

am

ag

ing

DA

MA

GIN

Gd

ise

ase

cau

sin

gD

AM

AG

ING

N.D

.eN

.D.

MM

Cf

Pro

toca

dh

eri

nre

pe

ats

c.3

37

2C

.G

NA

p.Il

e1

12

4M

et

po

ssib

lyd

am

ag

ing

DA

MA

GIN

Gd

ise

ase

cau

sin

gD

AM

AG

ING

N.D

.N

.D.

MM

CP

roto

cad

he

rin

rep

eat

s

c.4

08

5C

.T

NA

p.T

hr1

36

2M

et

pro

ba

bly

da

ma

gin

gD

AM

AG

ING

dis

ea

seca

usi

ng

DA

MA

GIN

GN

.D.

N.D

.M

MC

EGF-

like

(1–

3)

c.4

22

8G

.A

NA

p.G

ly1

41

0A

rgp

rob

ab

lyd

am

ag

ing

DA

MA

GIN

Gd

ise

ase

cau

sin

gT

OLE

RA

TED

N.D

.N

.D.

MM

CEG

F-lik

e(1

–3

)

c.4

92

7C

.T

NA

p.A

rg1

64

3T

rpb

en

ign

TO

LER

AT

EDp

oly

mo

rph

ism

TO

LER

AT

EDN

.D.

N.D

.M

MC

Lam

inin

G-l

ike

1

c.5

05

0_

50

51

ins

GT

NA

Tru

nca

ted

pro

tein

NA

NA

dis

ea

seca

usi

ng

NA

N.D

.N

.D.

MM

CEx

trac

ellu

lar

c.5

46

1G

.T

NA

p.V

al1

82

1Le

ub

en

ign

TO

LER

AT

EDp

oly

mo

rph

ism

TO

LER

AT

EDN

.D.

N.D

.M

MC

Lam

inin

G-l

ike

2

c.5

47

3G

.A

NA

p.G

ly1

82

5Se

rb

en

ign

TO

LER

AT

EDd

ise

ase

cau

sin

gD

AM

AG

ING

N.D

.N

.D.

MM

CLa

min

inG

-lik

e2

c.5

71

9_

57

20

de

lT

GN

AT

run

cate

dp

rote

inN

AN

Ad

ise

ase

cau

sin

gN

AN

.D.

N.D

.M

MC

Extr

ace

llula

r

c.6

18

4G

.A

NA

p.G

ly2

06

2Se

rp

oss

ibly

da

ma

gin

gT

OLE

RA

TED

dis

ea

seca

usi

ng

TO

LER

AT

EDN

.D.

N.D

.M

MC

Extr

ace

llula

r

c.7

06

0C

.T

NA

p.A

rg2

35

4C

ysp

rob

ab

lyd

am

ag

ing

DA

MA

GIN

Gp

oly

mo

rph

ism

DA

MA

GIN

GN

.D.

N.D

.M

MC

Extr

ace

llula

r

c.7

48

9C

.T

rs2

00

07

22

84

p.A

rg2

49

7C

ysp

rob

ab

lyd

am

ag

ing

DA

MA

GIN

Gd

ise

ase

cau

sin

gD

AM

AG

ING

N.D

.1

/10

00

MM

CT

ran

sme

mb

ran

e

c.8

63

2G

.A

NA

p.G

ly2

87

8Se

rb

en

ign

TO

LER

AT

EDp

oly

mo

rph

ism

TO

LER

AT

EDN

.D.

N.D

.M

MC

Cyt

op

lasm

icd

om

ain

aW

ech

ose

un

we

igh

ted

mo

de

lfo

ro

ur

mu

tati

on

fun

ctio

nal

eff

ect

pre

dic

tio

n;

bEx

om

eV

aria

nts

Pro

ject

dat

a(h

ttp

://e

vs.g

s.w

ash

ing

ton

.ed

u/E

VS/

);c1

00

0g

en

om

ese

qu

en

cin

gd

ata

(htt

p:/

/ww

w.1

00

0g

en

om

es.

org

/);

dN

Ast

and

sfo

rN

ot

avai

lab

le;

eN

.D.

ind

icat

es

No

td

ete

cte

d;

f MM

C:

mye

lom

en

ing

oce

le.

do

i:10

.13

71

/jo

urn

al.p

on

e.0

09

22

07

.t0

01

Association between CELSR1 Gene and Spina Bifida


http://evs.gs.washington.edu/EVS/

http://www.1000genomes.org/

lance system for collecting information on infants and fetuses with

congenital malformations, which has been described elsewhere

[21]. Included for study were 192 infants with isolated spina bifida

and without other major birth defects (cases) and 190 non-

malformed infants (controls). Cases randomly selected from all live

born infants with spina bifida and a random sample of non-

malformed infants were ascertained by the CBDMP correspond-

ing to birth years 1983–1999. Among the 192 spina bifida cases,

82 are White NonHispanic, 54 are native US born Hispanics and

56 are foreign born Hispanics. Among the 190 controls, 81 are

White NonHispanic, 54 are US born Hispanics and 55 are foreign

born Hispanics. All of the 192 spina bifida included in this study

are cases of myelomeningocele. The case and control infants were

linked to their newborn screening bloodspots, which served as the

source of the gDNA used in these studies. Bloodspots are collected

on all newborns in California for genetic testing purposes by the

State of California. The State retains the residual, unused portion

of the bloodspot and makes these bloodspots available to approved

researchers.

DNA sequencingGenomic DNA was extracted using the Puregene DNA

Extraction Kit (Qiagen, Valencia, CA) and amplified using the

GenomiPhi Kit (GE Healthcare). Coding exons and flanking

exon-intron regions of the human CELSR1 gene (NM_014246)

were amplified by polymerase chain reactions (PCR) from the

whole genome amplification (WGA) product. Primer sequences

are available upon request. The PCR products were sequenced

using the Prism Bigdye Terminator Kit (v3) on an ABI 3730XL

DNA analyzer (Life Technologies, Carlsbad, CA). Both case and

control samples were sequenced with either a specific forward or

reverse primer. Sequencing results were analyzed using the

Mutation Surveyor software V4.0.7 (Softgenetics, Stage College,

PA). Detected mutations were subsequently confirmed by a second

round of whole genome amplification, PCR and sequencing

analysis. GenomiPhi Kit takes advantage of Phi29 DNA

polymerase, which produces high fidelity during DNA replication

due to its proofreading 39–59 exonuclease activity. The reported

error rate of Phi29 is between 361026 [22] to 561026 [23]. The

probability to generate the same artifact mutation in two rounds of

WGA is 9610212 to 2.5610211. In our CELSR1 mutation screen,

Figure 1. TG dinucleotide repeats variants in spina bifida. A: Sequence trace of control (top) and C.5719–5720delTG (bottom). B: Sequencetrace of control (top) and TG duplication (bottom). C: Schematic representation of the CELSR1 predicted protein structure (accession numberQ9NYQ6) with the domains and approximate position of TG repeats variants.doi:10.1371/journal.pone.0092207.g001



we sequenced 192 spina bifida cases with 9045 nucleotides in the,

CELSR1(NM_014246) gene. In total, we screened 1.746106 base

pairs. The probability to detect the same coding region artifact

mutation in our study is less than 4.3561025.

PlasmidsMouse Ceslr1 cDNA cloned into a pEGFP-N1 plasmid

(pEGFPN1-Celsr1) was kindly provided to us by Dr. Elaine Fuchs

(The Rockefeller University, New York, USA). Celsr1 open reading

frame (ORF) was sub-cloned to pEGFP-C1. Human influenza

hemagglutinin (HA) tagged VANLG2 (HA-VANGL2) plasmid was

obtained from Dr. Hongyan Wang (Fudan University, Shanghai,

China). VANGL2 ORF was sub-cloned into the pDs-RedC1 at

XhoI and SalI restriction sites. CELSR1 nonsense and missense

changes were introduced into pEGFPN1-Celsr1 by QuikChange

II Site-Directed Mutagenesis Kits (Agilent Technologies, In-

c.CA,USA). All plasmids were validated by sequencing analyses.

Subcellular localizationMDCK II cells were purchased from Sigma-Aldrich and

cultured according to the manufacturer’s protocols. One day

before transfection, cells were seeded in 4 chamber 35 mm glass

bottom dishes (46104/chamber) (In Vitro Scientific, Sunnyvale,

CA). Plasmids transfection was performed using GeneTran III

Tranfection Reagent (Biomiga, San Diego, CA) according to the

manufacturer’s manual. Forty eight hours later, cells were washed

twice with PBS and incubated 5 minutes with Hochest 3342

(1 ug/ml) (Invitrogen), then washed 3 times with PBS and fixed in

4% PFA (paraformaldehyde in phosphate-buffered saline) for

10 minutes at 37uC, followed by 3 times PBS wash. Cells were

examined and photographed by an LSM710 laser scanning

confocal microscope (Leica).

Immunoprecipitation and immunoblottingHEK293T cells were grown and maintained in DMEM

supplemented with 10% fetal bovine serum (FBS) on a 6-well

plate at a concentration of 66105cell/well before the day of

transfection. 2 mg of GFP-Celsr1 or its related mutant plasmids

were co-transfected with 2 mg of HA-Vangl2 by Lipofectamine

reagent (Invitrogen). Twenty-four hours post-transfection, cells

were washed twice with ice-cold PBS and lysed with 300 ml lysis

buffer. Lysate was pretreated with protein A/G agarose, and then

immunoprecipitated with 1–2 mg anti-HA antibody and protein

A/G agarose at 4uC overnight. After washing three times with lysis

buffer, the precipitates were run on SDS-PAGE followed by

Western blot detection immunoblotting with the anti-GFP

antibody.

Results

Novel rare mutations identified in CELSR1 among spinabifida patients

The human CELSR1 gene coding region sequence contains 46

TGTG and 3 TGTGTG dinucleotide repeats (Figure S1). Two

Figure 2. Subcellular localization of GFP-Celsr1 wild type and TG repeat variants. A: MDCK II cells were transfected with GFP-Celsr1plasmids. Each image shown is representative of at least 50 examples. It demonstrated that the C.5719–5720delTG and the C.5050–5051insTG disruptGFP- Celsr1 membrane localization. Scale bar, 25 mm. B: Western blot of GFP-Celsr1 wild type and the two indels mutants. It showed that the c.5719–5720delTG and the c.5050–5051insTG changed the size of GFP-Celsr1 protein.doi:10.1371/journal.pone.0092207.g002



TG dinucleotide repeat variants were identified in spina bifida

cases (N = 192), and both of them were absent among the 190

control samples (Table 1 and Figure S2). One was a TG-insertion

(c.5050–5051insTG) and one was a TG-deletion (c.5719–

5720delTG), both of which created a stop codon in the middle

of the CELSR1 ORF. The C.5050–5051insTG created a stop

codon at the 1706th amino acid, and the C.5719–5720delTG

produced a stop codon at the 1944th amino acid (Figure 1). We

also identified 11 missense SNVs in NTD samples but not in any

controls, six of which were predicted to be deleterious or damaging

by both SIFT and PolyPhen (Table 1). The six mutations are

p.Arg2497Cys, p.Arg2354Cys, p.Gly1410Arg, p.Thr1362Met,

p.Ile1124Met and p.Ala1023Gly. Four of the 6 SNVs were

predicted to be damaging and disease causing by MutationTaster

and FATHMM, they were p.Arg2497Cys, p.Thr1362Met,

p.Ile1124Met and p.Ala1023Gly. These four mutations were

mapped to different domains of CELSR1: p.Ala1023Gly and

p.Ile1124 were mapped to the cadherin repeats, p.Thr1362Met

was mapped to a EGF-like domain and p.Arg2497Cye was

mapped to the transmembrane domain of CELSR1 (Table 1).

Among these four mutations, one (p.Arg2497Cys) was identified

once in the 1000 genome sequencing project. None of the 11

SNVs were previously detected in the Exome Variants Project

(Table 1) or the two previously published NTDs’ CELSR1

mutation screen studies [19,20]. All rare mutations identified in

this study have been uploaded to LOVD website (http://www.

lovd.nl/3.0/home).

The CELSR1 C.5050–5051insTG and the C.5719–5720delTG disrupt CELSR1 membrane localization

Several point mutations (p.Ala773Val, p.Arg2438Gln,

p,Ser2964Leu and p.Pro2983Ala) identified in humans craniora-

chischisis cases were found to alter membrane localization [19]. In

our study, both the C.5050–5051insTG and the C.5719–

5720delTG variantsintroduced stop codons ahead of the CELSR1

transmembrane domain. Therefore, we predicted that these two

variants would also impair CELSR1 membrane localization. In

the CELSR1 subcellular localization assay, both the C.5050–

5051insTG and the C.5719–5720delTG misplaced CELSR1 in

MDCK II cells. Unlike wild type CELSR1 which localizes almost

exclusively on the cell membrane, CELSR1 protein with C.5050–

5051insTG or C.5719–5720delTG distributed throughout the

cells, including the cytoplasm and the nucleus (Figure 2) in all of

the cells examined (100%).

The CELSR1 C.5050–5051insTG and the C.5719–5720delTG disrupt VANGL2 cell-contacts localization

It was demonstrated that Celsr1 is required to recruit Vangl2 to

sites of cell-cell contacts [24]. We subsequently studied whether the

Figure 3. Subcellular localization of Celsr1 and VANGL2 in MDCK II cells. A: The cells were transfected with GFP-Celsr1 or DsRed-VANGL2.GFP-Celsr1 itself was localized to cell membrane while DsRed-VANGL2 was unable to localize to cell membrane. B: The cells were co-transfected withdsRed-VANGL2 (Red) and GFP-Celsr1 wild type, C.5719–5720delTG, C.5050–5051insTG. In the presence of Celsr1WT–GFP, DsRed–Vangl2 changed itslocalization to cell–cell borders only when two Celsr1-expressing cells are in direct contact. Each image shown is representative of at least 30examples. It demonstrated that CELSR1 C.5050–5051insTG and C.5719–5720delTG mutants failed to recruit DsRed-VANGL2 to cell-cell contact.doi:10.1371/journal.pone.0092207.g003



http://www.lovd.nl/3.0/home

http://www.lovd.nl/3.0/home

C.5050–5051insTG and the C.5719–5720delTG affect the ability

of CELSR1 to recruit VANGL2 to the cell-cell contact. MDCK II

cells were transfected with PDS-RedC1-VANGL2 and GFP-

CELSR1 wild type or its variants. CELSR1 and VANGL2

subcellular localization were examined by confocal microscopy. As

shown in Figure 3, wild type CELSR1 co-localized with VANGL2

at the cell-cell contact in all of the cells examined (100%), while

TG indelmutant forms of CELSR1 could not recruit VANGL2 to

the contacting interface between co-transfected cells.

CELSR1 C.5050–5051insTG and C.5719–5720delTG impairinteraction with VANGL2

We also tested whether the CELSR1 C.5050–5051insTG and

the C.5719–5720delTG affect the interaction between CELSR1

and VANGL2. HEK293T cells were co-transfected with HA-

VANGL2 and either wild type or mutant GFP-CELSR1,

incubated for 24 hrs and protein lysates were subjected to co-

immunoprecipitation (Co-IP) and western blot analysis. The

expression levels of mutant and wild type GFP-CELSR1 were

found to be similar. Co-IP assay demonstrated that C.5050–

5051insTG and C.5719–5720delTG forms of CELSR1 impaired

Celsr1 interaction with HA-VANGL2 (Figure 4).

Discussion

Our study identified novel CELSR1 TG indels and SNV’s in

spina bifida patients. Two previous studies reported that rare

mutations in CELSR1 are associated with human NTDs [19,20].

Both previous studies identified only SNVs. One study investigated

the biological effects of some other SNV’s in detail, so our study

focused on evaluating the biological effect of the observed TG

repeat mutations.

Both the deletion and insertion were identified to be TG

dinucleotide repeats. The insertion c.5050–5051insTG added a

TG dinucleotide to TGTG, whereas c.5719–5720delTG removed

a TG dinucleotide repeat from TGTGTG. It is known that among

di-nucleotides, (TG)n are the most frequent in both humans and

mice [25]. In humans, CELSR1 coding sequence region (CDS) has

46 TGTG and 3 TGTGTG. In mice, Celsr1 CDS region has 63

TGTG and 6 TGTGTG repeats. In this California spina bifida

cohort, the CELSR1 CDS TG dinucleotide repeat variation rate

was 1% (2 in 192).

Previous studies showed that missense mutations in membrane

associated PCP genes including CELSR1 could affect PCP

pathway signaling by disrupting membrane localization [19].

Here, micro insertions/deletions created truncated CELSR1

forms lacking the transmembrane domain, so that membrane

localization might be prevented. Indeed, absence of mutant

CELSR1 from the membrane and its accumulation in the

cytoplasm and nuclear compartments was observed.

CELSR1 is known to physically associate with VANGL2, and

CELSR1-VANGL2 interaction plays an important role in

maintaining planar cell polarity during cell proliferation. During

mitosis, CELSR1 interacts with VANGL2 and recruits VANGL2

to endosomes. Following mitosis, PCP proteins are recycled to the

cell surface, where asymmetry is re-established by a process reliant

on neighboring PCP [16]. Mutations affecting CELSR1-VANGL2

interactions can preclude VANGL2 recruitment to the endosomes

during mitosis, thus disrupting PCP signaling. Both of the TG

dinucleotide repeat variants identified in this study prevented

Celsr1 physical association with VANGL2.

VANGL2 is another core PCP protein. It is a four-pass

transmembrane protein and its proper localization is critical for

VANGL2 function. Several proteins are required to establish

VANGL2 localization. Previous studies demonstrated that NTD-

inducing SNVs in Vangl2 itself, such as p.D255E and p.S464N,

could lead to mislocation of the protein in mice [26,27]. SEC24B

is a transport protein involved in vesicle trafficking, including

shaping of the vesicle, cargo selection and concentration.

Mutations of SEC24B affected VANGL2 membrane localization

[28,29]. Here, we demonstrated that truncation of CELSR1 can

disrupt VANGL2 cell-cell contact localization. Our results are

consistent with the findings in the previous study by Devenport

and Fuchs (2008) [24], which showed that deletion of the

cytoplasmic tail of GFP-Celsr1 partially impaired its localization

to contacting interfaces and its ability to recruit Vangl2 at these

cell-cell contacts. In our in vivo study, when GFP-Celsr1 wildtype

was transfected alone, Celsr1 was distributed uniformly at

membrane junctions (Figure 3A). When GFP-Celsr1 was co-

transfected with DsRedC1-VANGL2, it was distributed asymmet-

rically to cell-cell contacts (Figure 3B). This observation is

consistant with the Devenport and Fuchs (2008) study [24], which

indicated that Vangl2 and Celsr1 are dependent on one another

for their proper asymmetric distribution.

In summary, our study on a California spina bifida cohort

indicates that mutations in CELSR1 contribute to the development

of spina bifida. About 1% (2 in 192) of the spina bifida cases

presented TG indels in CELSR1. It is interesting that the two TG

indels identified in this study caused severe biological malfunction

of the mutant CELSR1 proteins, yet the birth defect associated

with them is myelomeningocele. In contrast, the study by

Robinson and co-workers [19] found SNV’s that moderately

altered the biological function of the CELSR1 protein, yet the

associated defect was craniorachischisis, a more severe type of

Figure 4. Effect of TG repeat variants on Celsr1-VANGL2interactions by Co-Ip. Western blotting of the lysate with anti-GFPdemonstrates the presence of wild-type GFP-Celsr1 protein and mutantproteins with different TG repeats. Following HA immunoprecipitation,blotting with anti-GFP confirmed physical association between Celsr1wild type and VANGL2, while this association was affected in CELSR1 TGrepeat variants.doi:10.1371/journal.pone.0092207.g004



NTD. One possible explanation is that NTDs are caused by a

combination of multiple genetic and environment factors. One

functional mutation is not sufficient to produce an NTD

phenotype. There likely needs to be multiple functional mutations

working together to cause an NTD phenotype, and to determine

the NTD subtype. It was demonstrated in mice that homozygous

PCP mutations caused craniochisichisis, as shown for Vangl2,

Celsr1, and Ptk7 while double heterozygosity for a PCP mutation

and non-PCP mutation can cause spina bifida, as shown for

Vangl2/Dact1 double heterozygotes. Perhaps the CELSR1 TG

indels combined with other non-PCP mutations caused the

myelomeningoceles that we observed, and that in the Robinson’s

study, the moderate CELSR1 mutations combined with other

functional PCP mutations resulted in craniorachischisis. In our

study, none of the CELSR1 TG indels were combined with

mutations in other sequenced PCP genes including VANGL1,

VANGL2, DISHEVELLED1,DISHEVELLED2, DISHEV-

ELLED3, FZD6, SCRIB and PTK7. Exome sequencing or whole

genome sequencing approaches are warranted in order to examine

the role of additional PCP genes such as PRICKLE, ANKRD6

(ankyrin repeat domain 6; also known as DIVERSIN, the

orthologue of diego), FUZ, and to detect non-PCP genes that

may contribute to these NTD phenotypes.

The four SNVs, which were predicted to be damaging by the

four computational programs used in this study, were mapped to

three different domains of CELSR1. Two of them, p.Ala1023Gly

and p.Ile1124Met were mapped to the eighth cadherin repeat

domain of CELSR1, the same domain where two previously

identified NTD mouse mutations (p.Asp1040Gly and

p.Asp1110Lys) were located. These two SNVs may affect the

function of cadherin repeat domain which plays an important role

in cell-cell contact. Variant p.Thr1362Met was predicted to

change a threonine amino acid to a methionine amino acid.

Thr1362 is a putative phosphorylation site of both protein kinase

B (PKB) and G protein-coupled receptor kinase (GRK) based on

analysis using KinasePhos 2.0 (http://kinasephos2.mbc.nctu.edu.

tw/). Variant p.Arg2497Cys was mapped to the seven transmem-

brane domain. The amino acid characteristics change from

arginine’s basic to cysteine’s hydroxyl may affect protein structure

and CELSR1 membrane localization.

Combined with the SNVs, about 3% (6 in 192) of the spina

bifida patients in our cohort possess CELSR1 deleterious or

predicted-to-be-deleterious variants. Our data provides further

evidence emphasizing the contributions of PCP genes to the

etiology of NTDs.

Supporting Information

Figure S1 The 46 TGTG and 3 TGTGTG in humanCELSR1 (NM_014246) coding region sequence. TGTG

repeats were highlighted by green color and TGTGTG repeats

were highlighted by red color. Both of the two TG repeat variants

(c.5050–5051insTG and c.5719–5720delTG) were underlined.

(DOCX)

Figure S2 Electropherograms of CELSR1 TG indelsfrom genomic DNA. Panel A indicated forward and reverse

primer sequencing result of c.5719–5720del TG. Panel B indicated

forward and reverse primer sequencing result of c.5050–5051ins

TG.

(DOCX)

Acknowledgments

We thank the California Department of Public Health, Maternal Child and

Adolescent Health Division for providing data. We thank Dr. Elaine Fuchs

of the Rockefeller University for providing the pEGFPN1-Celsr1 plasmid.

We also appreciate the outstanding confocal microscopy support provided

by Dr. Yue Li of the Dell Pediatric Research Institute. The findings and

conclusions in this report are those of the authors and do not necessarily

represent the official position of the California Department of Public

Health.

Author Contributions

Conceived and designed the experiments: RHF. Performed the experi-

ments: YL. Analyzed the data: YL. Contributed reagents/materials/

analysis tools: WY GMS. Wrote the paper: YL HZ MER GMS RHF.

References

1. Group MVSR (1991) Prevention of neural tube defects: results of the Medical

Research Council Vitamin Study. Lancet 338: 131–137.

2. Deak KL, Siegel DG, George TM, Gregory S, Ashley-Koch A, et al. (2008)

Further evidence for a maternal genetic effect and a sex-influenced effect

contributing to risk for human neural tube defects. Birth Defects Res A Clin Mol

Teratol 82: 662–669.

3. Shaw GM, Lu W, Zhu H, Yang W, Briggs FB, et al. (2009) 118 SNPs of folate-

related genes and risks of spina bifida and conotruncal heart defects. BMC Med

Genet 10: 49.

4. Olshan AF, Shaw GM, Millikan RC, Laurent C, Finnell RH (2005)

Polymorphisms in DNA repair genes as risk factors for spina bifida and

orofacial clefts. Am J Med Genet A 135: 268–273.

5. Tran PX, Au KS, Morrison AC, Fletcher JM, Ostermaier KK, et al. (2011)

Association of retinoic acid receptor genes with meningomyelocele. Birth Defects

Res A Clin Mol Teratol 91: 39–43.

6. Juriloff DM, Harris MJ (2012) A consideration of the evidence that genetic

defects in planar cell polarity contribute to the etiology of human neural tube

defects. Birth Defects Res A Clin Mol Teratol 94: 824–840.

7. Simons M, Mlodzik M (2008) Planar cell polarity signaling: from fly

development to human disease. Annu Rev Genet 42: 517–540.

8. Ybot-Gonzalez P, Savery D, Gerrelli D, Signore M, Mitchell CE, et al. (2007)

Convergent extension, planar-cell-polarity signalling and initiation of mouse

neural tube closure. Development 134: 789–799.

9. Greene ND, Stanier P, Copp AJ (2009) Genetics of human neural tube defects.

Hum Mol Genet 18: R113–129.

10. De Marco P, Merello E, Rossi A, Piatelli G, Cama A, et al. (2012) FZD6 is a

novel gene for human neural tube defects. Hum Mutat 33: 384–390.

11. De Marco P, Merello E, Consales A, Piatelli G, Cama A, et al. (2013) Genetic

analysis of disheveled 2 and disheveled 3 in human neural tube defects. J Mol

Neurosci 49: 582–588.

12. Kibar Z, Torban E, McDearmid JR, Reynolds A, Berghout J, et al. (2007)Mutations in VANGL1 associated with neural-tube defects. N Engl J Med 356:

1432–1437.

13. Lei YP, Zhang T, Li H, Wu BL, Jin L, et al. (2010) VANGL2 mutations in

human cranial neural-tube defects. N Engl J Med 362: 2232–2235.

14. Lei Y, Zhu H, Duhon C, Yang W, Ross ME, et al. (2013) Mutations in planarcell polarity gene SCRIB are associated with spina bifida. PLoS One 8: e69262.

15. Bosoi CM, Capra V, Allache R, Trinh VQ, De Marco P, et al. (2011)Identification and characterization of novel rare mutations in the planar cell

polarity gene PRICKLE1 in human neural tube defects. Hum Mutat 32: 1371–1375.

16. Devenport D, Oristian D, Heller E, Fuchs E (2011) Mitotic internalization of

planar cell polarity proteins preserves tissue polarity. Nat Cell Biol 13: 893–902.

17. Formstone CJ, Mason I (2005) Combinatorial activity of Flamingo proteins

directs convergence and extension within the early zebrafish embryo via theplanar cell polarity pathway. Dev Biol 282: 320–335.

18. Curtin JA, Quint E, Tsipouri V, Arkell RM, Cattanach B, et al. (2003) Mutation

of Celsr1 disrupts planar polarity of inner ear hair cells and causes severe neuraltube defects in the mouse. Curr Biol 13: 1129–1133.

19. Robinson A, Escuin S, Doudney K, Vekemans M, Stevenson RE, et al. (2012)Mutations in the planar cell polarity genes CELSR1 and SCRIB are associated

with the severe neural tube defect craniorachischisis. Hum Mutat 33: 440–447.

20. Allache R, De Marco P, Merello E, Capra V, Kibar Z (2012) Role of the planarcell polarity gene CELSR1 in neural tube defects and caudal agenesis. Birth

Defects Res A Clin Mol Teratol 94: 176–181.

21. Croen LA, Shaw GM, Jensvold NG, Harris JA (1991) Birth defects monitoring

in California: a resource for epidemiological research. Paediatr PerinatEpidemiol 5: 423–427.

22. Nelson JR, Cai YC, Giesler TL, Farchaus JW, Sundaram ST, et al. (2002)

TempliPhi, phi29 DNA polymerase based rolling circle amplification of

templates for DNA sequencing. Biotechniques Suppl: 44–47.



http://kinasephos2.mbc.nctu.edu.tw/

http://kinasephos2.mbc.nctu.edu.tw/

23. Esteban JA, Salas M, Blanco L (1993) Fidelity of phi 29 DNA polymerase.

Comparison between protein-primed initiation and DNA polymerization. J BiolChem 268: 2719–2726.

24. Devenport D, Fuchs E (2008) Planar polarization in embryonic epidermis

orchestrates global asymmetric morphogenesis of hair follicles. Nat Cell Biol 10:1257–1268.

25. Ellegren H (2004) Microsatellites: simple sequences with complex evolution. NatRev Genet 5: 435–445.

26. Kibar Z, Vogan KJ, Groulx N, Justice MJ, Underhill DA, et al. (2001) Ltap, a

mammalian homolog of Drosophila Strabismus/Van Gogh, is altered in themouse neural tube mutant Loop-tail. Nat Genet 28: 251–255.

27. Murdoch JN, Doudney K, Paternotte C, Copp AJ, Stanier P (2001) Severe

neural tube defects in the loop-tail mouse result from mutation of Lpp1, a novel

gene involved in floor plate specification. Hum Mol Genet 10: 2593–2601.

28. Merte J, Jensen D, Wright K, Sarsfield S, Wang Y, et al. (2010) Sec24b

selectively sorts Vangl2 to regulate planar cell polarity during neural tube

closure. Nat Cell Biol 12: 41–46; sup pp 41–48.

29. Yang XY, Zhou XY, Wang QQ, Li H, Chen Y, et al. (2013) Mutations in the

COPII Vesicle Component Gene SEC24B are Associated with Human Neural

Tube Defects. Hum Mutat.



identification of novel celsr1 mutations in spina bifida

Documents